Geographical Pattern of VOT in Japanese Initial Voiced Stops

Abstract

This research studied Voice Onset Time (VOT) in Japanese word-initial voiced stops, with the purpose of finding patterns of geographical variation. Although it is generally believed that Japanese voiced stops basically take negative values, my previous studies (Takada 2006, 2007a, 2007b) show that there are (1) regional differences, (2) generational differences, and (3) no co-occurrence of nasalization and voicing in the case of medial stops. The entire picture of the VOT's geographical pattern, however, is yet unknown. Data for the acoustic study include a speech corpus that was collected throughout Japan in the years 1986-1988, and recent recordings, made in the years 2006-2007, that additionally include some yet unresearched areas. In total, about 450 speakers pronounced 15 words with an initial voiced stop. VOT was analyzed with the help of a waveform and spectrogram. General results suggest a clear geographical pattern as follows. Older generation speakers show an East-West regional opposition, one of the most common patterns in Japanese dialectology. Namely, VOT values for Tohoku (East regional variety) evenly match positive values, whereas on the other hand, Kanto and areas west (West regional variety) are unified by negative VOT values. The latter variety is suggested as having a peripheral distribution, of which the Kinki area is thought to be most typical. These regional differences, however, are found to be unclear in younger generations. The reason for this is thought to be changes in the West regional variety, wherein VOT values are getting larger, and gradually turning into positive ones.